Skip to content

README.md

Latest commit

 

History

History
72 lines (54 loc) · 3.06 KB

README.md

File metadata and controls

72 lines (54 loc) · 3.06 KB
Raw

Mulled Container Builder Website

This static website lets users browse bioconda and conda-forge packages, assemble a basket of name=version pairs, and open a pull request against combinations/hash.tsv with a single click.

It is built with Astro + Tailwind CSS and deployed to the gh-pages branch by CI. A daily cron rebuild keeps the package index up to date with the latest conda repodata.

Local development

just install    # install npm dependencies
just dev        # local dev server with hot reload
just build      # fetch conda repodata + build the static site
just build-offline  # build without network (empty package index)
just test       # build with test fixtures + run Playwright tests

How it works

  1. Build time: website/scripts/fetch_repodata.mjs fetches repodata.json from conda.anaconda.org for both bioconda and conda-forge, extracts package names + all versions, and writes compact JSON to website/public/data/. It also reads all combinations/*.tsv files (including hash.tsv) to produce combinations.json and used-names.json.

  2. Runtime: The static site fetches these JSON files lazily. Search, basket, and existing-combinations browser are client-side islands. The "Build container" button generates a GitHub web editor URL that pre-fills a new combination file — no token or backend needed.

  3. Deployment: The .github/workflows/website.yml workflow builds the site and pushes website/dist/ to the gh-pages branch. It runs on every push to master and on a daily cron schedule.

Data API

The following JSON files are generated at build time and served as static assets from the website. They are accessible at:

File Description
/data/packages.json All bioconda and conda-forge packages with all versions: {"bioconda": [{"name": "samtools", "versions": ["1.20", ...]}], "conda-forge": [...]}
/data/combinations.json All existing combinations from combinations/*.tsv (including hash.tsv): [{"targets": "samtools=1.20", "base_image": "", "image_build": "0", "file": "hash.tsv"}]
/data/used-names.json Sorted array of package names that already appear in at least one existing combination: ["samtools", "bwa", ...]

How PRs are created

No GitHub token or backend API is needed. The website builds a URL to GitHub's web editor:

https://github.com/{owner}/{repo}/new/master
  ?filename=combinations/{slug}.tsv
  &value={url-encoded TSV content}

When the user clicks "Build container", this URL opens in a new tab with:

  • The filename pre-filled as combinations/{targets-slug}.tsv
  • The content pre-filled as a TSV with the header and the selected name=version targets

The user clicks "Propose changes" in GitHub's UI, which creates a branch and opens a pull request. The existing CI then builds and publishes the container once the PR is merged.

Using your own fork

To open PRs against your own fork, change the Owner and Repository fields in the "Build container" panel on the website.